Using Correlations to Call Turns in NVDA/S&P 500 (Transcript)
By datatrekresearch in Blog
(Published August 14, 2024)
DataTrek co-founder Jessica Rabe shares one key measure we’ve used to make the right calls on US equities this year, including the recent selloff. One example: S&P 500 sector correlations to the index were abnormally low in mid-July, which told us that investors were too optimistic about US stocks. The pullback since has seen correlations increase back to near-normal levels. We’ll want to see them stabilize here to confirm the bull market is still alive and well.
Watch it here on our YouTube channel! Please hit the like and subscribe button if you find it useful, thank you.
Transcript:
Hi, I’m Jessica Rabe, one of the co-founders of DataTrek Research. With all the recent stock market volatility, today’s video is will show you how we use correlation analysis to assess investor sentiment and future returns.
I’ll do this through two examples. The first will look at S&P sector correlations to the market, and the second will focus on Nvidia, a stock I have owned personally for several years.
But before we get into the data, I’ll briefly explain correlations, which I know is a super nerdy topic, but as I walk you through the examples you’ll see why it’s such an important topic. Simply put, correlations measure how closely two variables track each other over time. They run on a scale from one, which is a perfect positive correlation, to negative one, which is a perfect negative correlation. A zero correlation means there’s no statistical relationship and that the two variables move independently of one another.
With that, let’s dive into our first example. This chart shows trailing 30-day average price return correlations between 5 major large cap sectors and the S&P 500 itself back to 2018. These 5 sectors make up nearly three quarters of the S&P and are Tech, Health Care, Consumer Discretionary, Industrials, and Financials.The solid red line at 0.81 shows the average sector correlation to the S&P over this period, and the dotted lines show the 1 standard deviation band. Any time the blue line is above or below those dotted lines, you know sector correlations are unusually high or low.
I’ll first describe what happens when correlations are abnormally high or low, and then show how we used this measure to help inform our bullish call on equities for most of this year and then why we warned our clients last month about near-term choppiness, which ended up happening right on cue.
When macro uncertainty is very high, average S&P sector correlations always approach one – a near perfect positive correlation – because investors reduce equity exposure across the board: in other words, everything gets sold. You can see towards the left side of the chart where we marked that sector correlations were above 0.9 at the end of 2018 because of Chair Powell’s policy mistake, again in March 2020 at the peak of the Pandemic Crisis, and again towards the right of the chart in 2022 amid rapidly rising rates.
Now, it’s important to remember that correlations have an inverse relationship with performance, meaning that when they go up, stock prices go down. That was obviously the case in late 2018 until Chair Powell capitulated in early 2019, also in early 2020 at the onset of the public health crisis, and again in 2022 as the Fed hiked rates to fight inflation.
On the other end of the spectrum, correlations tend to be lower than usual when investors are confident in future economic and market conditions, which encourages them to choose between potential winners and losers. For example, correlations were as low as 0.6 in 2020 and 2021 as US equities rebounded from the short pandemic-driven bear market. That’s because massive fiscal and monetary policy stimulus alleviated investors’ uncertainty and gave them confidence that market and economic conditions would improve.
As for what S&P sector correlations can tell us about the current environment, consider that they fell to a 6-year low of 0.38 on July 12th, which was 4 standard deviations below the mean. That’s an extremely unusual level and told us that investors were way too optimistic about the macro environment and, therefore, future stock market returns. It’s therefore no coincidence that the S&P 500 hit its all-time high two trading days later. At that point, the S&P was up 18.8 pct YTD through its record high on July 16th, two trading days after the 6-year low in sector correlations because again, returns move oppositely of correlations, so equities rally as correlations fall.
Additionally, such low sector correlations helped explain July’s powerful rotation into US small caps. That’s because it took a lot of investor confidence to switch out of the market’s largest outperformers since early 2023, namely Big Tech, in favor a big underperformer, in this case small caps.
In terms of the recent selloff, sector correlations moved up sharply from a very unusual low of 0.38 on the day of the S&P’s record high on July 16th to 0.69 on the day of the S&P’s bottom so far on August 5th. In other words, S&P sector correlations almost doubled over 14 trading days, while the index declined by 8.5 pct over the same time frame. At the time of recording this video, the S&P has since rallied (?) by nearly 5 pct, while sector correlations have kept creeping higher back towards the long run average of 0.81, but that’s because our timeframe captures the trailing 30 days of market action. As US equities keep rallying, sector correlations should remain steady or fall once again as long as there’s no systemic shocks and the Fed delivers on rate cuts. Even still, this data does not yet show that inflection point and says it’s still too early to confidently say the coast is clear.
Let’s move on now to our second example. This chart shows the trailing 30-day price return correlations between Nvidia and the S&P since the start of 2018 in the same format as the prior graph:
Nvidia’s average correlation to the index is 0.65, but its 30-day correlation fell to as low as 0.1 on June 18th. That was almost 3 standard deviations below the mean and therefore super rare. Nvidia completely broke away from the US stock market and became essentially its own market.
Now, not only did June 18th mark the low in Nvidia’s correlation to the S&P, but it was also the date of Nvidia’s record close. Nvidia was up a remarkable 131% YTD through that all time high. Again, it is no coincidence that an abnormally low correlation happened concurrently with a record close. Investors were extremely confident in the stock in mid-June, and clearly too much so.
With the dramatic rotation out of US Big Tech and into small caps in July, and then subsequent broader US equity market selloff during the first week of August, Nvidia’s stock price declined 27% from its record high in mid-June to its bottom so far on August 7th. Over that same timeframe, Nvidia’s correlation to the S&P spiked from 0.1 to 0.74, well above the average of 0.65. Even though that move was disappointing, it was actually a healthy development because correlations should never have gotten as low as they did. At the time of this recording, Nvidia has recovered 17 pct since. The stock’s correlation to the S&P has continued to trend higher to 0.77, but it’s still within one standard deviation of the mean and therefore normal. At the same time, we’ll want to see this correlation start moving lower in the coming weeks to show that Nvidia’s current rebound has legs to it.
With that, thank you very much for taking the time to watch this video, please feel free to share it with anyone you think may find it useful. And as always, please hit the like button and subscribe to our YouTube channel if you enjoyed it.You can check out our website datatrekresearch.com for a free trial to our daily reports on markets, data and disruption. There’s no credit card or personal information required, just drop your email in.
Thanks again for watching and we hope you have a great day!
